[gentoo-dev] Patches and the 2.4->2.6 move

[gentoo-dev] Patches and the 2.4->2.6 move

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 795 bytes --]

I'd like some input on this:

I have a patch that fixes SSE problems on 2.4 kernels. However, on 2.6
kernels it breaks OpenGL (applications segfault).

Currently I'm applying this patch if /usr/src/linux is linked to a 2.4
kernel at compile-time.

This means that if a user emerges xfree when linked to a 2.4 kernel,
that user will need to remerge xfree after moving to 2.6 kernels. This
takes about 40 minutes on a ~2GHz x86. However, if this patch is not
applied, the bug will continue to exist for all Gentoo users on 2.4
kernels.

My request to you is:
1) Is this acceptable?
2) If not, what is a better solution? Dropping the patch entirely? Note
that I don't know Mesa/programming well enough to write a patch
compatible for both 2.4 and 2.6 kernels.

Thanks,
Donnie

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Luke-Jr]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 20 October 2003 02:34 am, Donnie Berkholz wrote:
> I have a patch that fixes SSE problems on 2.4 kernels. However, on 2.6
> kernels it breaks OpenGL (applications segfault).
Shouldn't this be fixed in the kernel, then? What exactly does it actually 
fix?
>
> Currently I'm applying this patch if /usr/src/linux is linked to a 2.4
> kernel at compile-time.
>
> This means that if a user emerges xfree when linked to a 2.4 kernel,
> that user will need to remerge xfree after moving to 2.6 kernels. This
> takes about 40 minutes on a ~2GHz x86. However, if this patch is not
x86 is not specific enough to give 2GHz any meaning. A 2GHz AthlonXP is alot 
faster than an (early) 2GHz Pentium 4.
> applied, the bug will continue to exist for all Gentoo users on 2.4
> kernels.
>
> My request to you is:
> 1) Is this acceptable?
40 minutes on any kind of 2GHz system would probably mean at least 2 hours for 
me... I'd recommend at least having a local USE flag or variable to enable/
disable it (default depending on how major the fix is)
> 2) If not, what is a better solution? Dropping the patch entirely? Note
> that I don't know Mesa/programming well enough to write a patch
> compatible for both 2.4 and 2.6 kernels.
If a version compatible with both kernels could exist, perhaps asking in some 
related IRC channels could find someone interested in writing such?
>
> Thanks,
> Donnie
- -- 
Luke-Jr
Developer, Gentoo Linux
http://www.gentoo.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQE/k1EbZl/BHdU+lYMRAnFdAJ4ujALMO8sUVOyMSp3AZimOQF6iCwCfSGzd
5iJIil8aFriglB3ZT0YEyPY=
=FXhs
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 1875 bytes --]

On Sun, 2003-10-19 at 23:05, Luke-Jr wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Monday 20 October 2003 02:34 am, Donnie Berkholz wrote:
> > I have a patch that fixes SSE problems on 2.4 kernels. However, on 2.6
> > kernels it breaks OpenGL (applications segfault).
> Shouldn't this be fixed in the kernel, then? What exactly does it actually 
> fix?
> >
> > Currently I'm applying this patch if /usr/src/linux is linked to a 2.4
> > kernel at compile-time.
> >
> > This means that if a user emerges xfree when linked to a 2.4 kernel,
> > that user will need to remerge xfree after moving to 2.6 kernels. This
> > takes about 40 minutes on a ‾2GHz x86. However, if this patch is not
> x86 is not specific enough to give 2GHz any meaning. A 2GHz AthlonXP is alot 
> faster than an (early) 2GHz Pentium 4.
> > applied, the bug will continue to exist for all Gentoo users on 2.4
> > kernels.
> >
> > My request to you is:
> > 1) Is this acceptable?
> 40 minutes on any kind of 2GHz system would probably mean at least 2 hours for 
> me... I'd recommend at least having a local USE flag or variable to enable/
> disable it (default depending on how major the fix is)

A USE flag to disable fixing a bug?
'USE="keep-sse-bugs" emerge xfree' seems a little odd to me.

> > 2) If not, what is a better solution? Dropping the patch entirely? Note
> > that I don't know Mesa/programming well enough to write a patch
> > compatible for both 2.4 and 2.6 kernels.
> If a version compatible with both kernels could exist, perhaps asking in some 
> related IRC channels could find someone interested in writing such?

I already tried. Response consisted of things similar to "Wait until 2.6
is official, then we will provide a fix for it. Right now we have more
important, relevant things to concentrate on."

Thanks,
Donnie

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 539 bytes --]

On Sun, 2003-10-19 at 23:05, Luke-Jr wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Monday 20 October 2003 02:34 am, Donnie Berkholz wrote:
> > I have a patch that fixes SSE problems on 2.4 kernels. However, on 2.6
> > kernels it breaks OpenGL (applications segfault).
> Shouldn't this be fixed in the kernel, then? What exactly does it actually 
> fix?

Forgot to answer this:
"Fix NASM problems, remove static var in SSE detection code.  (Josh
Vanderhoof)"

It fixes problems in SSE and 3Dnow detection.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Luke-Jr]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 20 October 2003 03:22 am, Donnie Berkholz wrote:
> Forgot to answer this:
> "Fix NASM problems, remove static var in SSE detection code.  (Josh
> Vanderhoof)"
>
> It fixes problems in SSE and 3Dnow detection.
But why doesn't it work the same with both 2.4 and 2.6? Is it a bug in 2.4 or 
2.6 that makes them react differently/poorly?
- -- 
Luke-Jr
Developer, Gentoo Linux
http://www.gentoo.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQE/k17fZl/BHdU+lYMRAsw3AJ0bzxsQ6B0DhrUgTOP/27sPhDd/cgCcC8Z+
/j3K7zfj22cO4CBBKDWvCh4=
=B3EZ
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 541 bytes --]

On Mon, 2003-10-20 at 00:04, Luke-Jr wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Monday 20 October 2003 03:22 am, Donnie Berkholz wrote:
> > Forgot to answer this:
> > "Fix NASM problems, remove static var in SSE detection code.  (Josh
> > Vanderhoof)"
> >
> > It fixes problems in SSE and 3Dnow detection.
> But why doesn't it work the same with both 2.4 and 2.6? Is it a bug in 2.4 or 
> 2.6 that makes them react differently/poorly?

If I knew that, I could probably write a patch that applied to both.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Brandon Hale]

> I have a patch that fixes SSE problems on 2.4 kernels. However, on 2.6
> kernels it breaks OpenGL (applications segfault).
>
> Currently I'm applying this patch if /usr/src/linux is linked to a 2.4
> kernel at compile-time.
>
> This means that if a user emerges xfree when linked to a 2.4 kernel,
> that user will need to remerge xfree after moving to 2.6 kernels. This
> takes about 40 minutes on a ~2GHz x86. However, if this patch is not
> applied, the bug will continue to exist for all Gentoo users on 2.4
> kernels.

It appears that the larger part of our user base is running a 2.4 series kernel,
and will be for quite some time. Considering this, and our position that
testing kernels are not offically supported, I think we should give priority to
fixing bugs that affect the most users.

> My request to you is:
> 1) Is this acceptable?

I find this acceptable under the condition that it is properly documented in
both postinst and ChangeLog.

> 2) If not, what is a better solution? Dropping the patch entirely? Note
> that I don't know Mesa/programming well enough to write a patch
> compatible for both 2.4 and 2.6 kernels.

The one caveat I can think of is GRP. Will we distribute a package for both
kernels? How does portage choose the right one?

> Thanks,
> Donnie

--tseng

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 434 bytes --]

On Sun, 2003-10-19 at 23:20, Brandon Hale wrote:
> The one caveat I can think of is GRP. Will we distribute a package for both
> kernels? How does portage choose the right one?

I say distribute whatever matches the kernel sources on the CD (2.4 or
2.6). Since XFree will have been compiled against those sources anyway
(whatever /usr/src/linux is in the chroot) this should be seamless,
based on my knowledge of such things.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Luke-Jr]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 20 October 2003 03:25 am, Donnie Berkholz wrote:
> I say distribute whatever matches the kernel sources on the CD (2.4 or
> 2.6). Since XFree will have been compiled against those sources anyway
> (whatever /usr/src/linux is in the chroot) this should be seamless,
> based on my knowledge of such things.
The CD could contain both kernels. A patch from 2.4.21 to 2.6.0-test7 is only 
8.2 MB (though you might as well stick the entire 2.6 test on there as it's 
only 30-some MB)
- -- 
Luke-Jr
Developer, Gentoo Linux
http://www.gentoo.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQE/k2VtZl/BHdU+lYMRAq23AJ9hYInT2V3UbMBJLznMnooGsdXTGACdG7Rg
N265mw6gLKPPbft0mD8TfZc=
=NgJv
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Luke-Jr]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 20 October 2003 03:20 am, Brandon Hale wrote:
> It appears that the larger part of our user base is running a 2.4 series
> kernel, and will be for quite some time. Considering this, and our position
> that testing kernels are not offically supported, I think we should give
> priority to fixing bugs that affect the most users.
I was thinking that once 2.6.0 was final, the kernel people would (hopefully) 
release a gentoo-sources using 2.6 in unstable... If this is not the case, 
would anyone mind if love-sources were to be added to the portage tree or is 
there any reason not to?
- -- 
Luke-Jr
Developer, Gentoo Linux
http://www.gentoo.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQE/k2GMZl/BHdU+lYMRAow0AJ4qOy2QdLiSGBAHK/MbNw0qm1on4wCdG5lS
vDFXTcyyDWNgYx2LJ2EaxPM=
=1x5y
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Jason Stubbs]

On Monday 20 October 2003 13:16, Luke-Jr wrote:
> I was thinking that once 2.6.0 was final, the kernel people would
> (hopefully) release a gentoo-sources using 2.6 in unstable... If this is
> not the case, would anyone mind if love-sources were to be added to the
> portage tree or is there any reason not to?

love-sources is the work of two people who both have fairly little time to 
devote to the numerous bug reports coming from the relatively small number of 
users using it. If it were added to portage (even if it was masked in 
packages.mask) the number of bug reports would be ridiculous. I suggest that 
it should not be added unless it becomes more than just a hobby for the 
maintainers.

Regards,
Jason

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Jon Portnoy]

On Mon, Oct 20, 2003 at 04:16:08AM +0000, Luke-Jr wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Monday 20 October 2003 03:20 am, Brandon Hale wrote:
> > It appears that the larger part of our user base is running a 2.4 series
> > kernel, and will be for quite some time. Considering this, and our position
> > that testing kernels are not offically supported, I think we should give
> > priority to fixing bugs that affect the most users.
> I was thinking that once 2.6.0 was final, the kernel people would (hopefully) 
> release a gentoo-sources using 2.6 in unstable... If this is not the case, 
> would anyone mind if love-sources were to be added to the portage tree or is 
> there any reason not to?

Yes.

The last thing we need is more kernel sources in the tree, especially 
from amateurs.

The kernel people will get a 2.6 kernel out when it's ready to be used, 
I'd assume. 

-- 
Jon Portnoy
avenj/irc.freenode.net

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Kumba]

Donnie Berkholz wrote:

> I'd like some input on this:
> 
> I have a patch that fixes SSE problems on 2.4 kernels. However, on 2.6
> kernels it breaks OpenGL (applications segfault).
> 
> Currently I'm applying this patch if /usr/src/linux is linked to a 2.4
> kernel at compile-time.
> 
> This means that if a user emerges xfree when linked to a 2.4 kernel,
> that user will need to remerge xfree after moving to 2.6 kernels. This
> takes about 40 minutes on a ~2GHz x86. However, if this patch is not
> applied, the bug will continue to exist for all Gentoo users on 2.4
> kernels.
> 
> My request to you is:
> 1) Is this acceptable?
> 2) If not, what is a better solution? Dropping the patch entirely? Note
> that I don't know Mesa/programming well enough to write a patch
> compatible for both 2.4 and 2.6 kernels.
> 
> Thanks,
> Donnie


Here's an idea.  Bit demented, though.

First, why check /usr/src/linux?  IMHO, that link may or may not always 
be there or be correct.  I recommend instead using the get_KV and 
KV_to_int functions defined in /sbin/functions.sh to determine what the 
machine is currently running.

The second part is where to apply the patch, and how.  This is the 
difficult part.  I'd first say to apply it from the kernel.eclass file, 
but this likely goes against some odd policy somewheres, which means 
most modern 2.4 ebuilds would need updating to use this patch.  Alot of 
work is involved in tweaking all the 2.4 sources ebuilds.

The third part is How to know when to apply the patch?  In pseudo-codish 
form, here is what I can come up with.

if KV_to_int(get_KV) <= KV_to_int(2.4.99) then
	if (use x86) && (use xfree) then
		epatch <patch>
		echo "${BAD}*${NORMAL} "
		echo "${BAD}*${NORMAL} BIG FAT WARNING: You are running Xfree on an 
x86  system and are running a 2.4 kernel.
		echo "${BAD}*${NORMAL} Because of this, a patch has been applied to 
the kernel to fix some SSE/3DNOW problems."
		echo "${BAD}*${NORMAL} This patch requires you to rebuild Xfree in 
order to take advantage of the patch's"
		echo "${BAD}*${NORMAL} benefits.  Please note that this patch is only 
for 2.4 kernels.  If you move to a 2.6"
		echo "${BAD}*${NORMAL} kernel, you WILL need to re-merge Xfree, 
otherwise OpenGL-based applications"
		echo "${BAD}*${NORMAL} will fail in X."
		echo "${BAD}*${NORMAL} "
	fi
fi


That should go into pkg_postinst, and probably include a sleep delay 
timer + bell dinging like on portage and baselayout ebuilds to get the 
attention of users.

All speculation, of course, but it's a slow night and I'm in brainstorm 
mode.


--Kumba

-- 
"Such is oft the course of deeds that move the wheels of the world: 
small hands do them because they must, while the eyes of the great are 
elsewhere."  --Elrond


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 1605 bytes --]

On Mon, 2003-10-20 at 00:33, Kumba wrote:
> Donnie Berkholz wrote:
<snip>
> > Currently I'm applying this patch if /usr/src/linux is linked to a 2.4
> > kernel at compile-time.
<snip>
> Here's an idea.  Bit demented, though.
> 
> First, why check /usr/src/linux?  IMHO, that link may or may not always 
> be there or be correct.  I recommend instead using the get_KV and 
> KV_to_int functions defined in /sbin/functions.sh to determine what the 
> machine is currently running.

Because everything in Gentoo that needs a kernel should compile against
/usr/src/linux. pcmcia-cs does so, xfree-drm does so, nvidia-kernel does
so, ati-drivers does so ...

> 
> The second part is where to apply the patch, and how.  This is the 
> difficult part.  I'd first say to apply it from the kernel.eclass file, 
> but this likely goes against some odd policy somewheres, which means 
> most modern 2.4 ebuilds would need updating to use this patch.  Alot of 
> work is involved in tweaking all the 2.4 sources ebuilds.

Hm, I think you may be misunderstanding here. The patch is against
xfree, but based on which kernel is there, so applying from a kernel
eclass doesn't really make sense.

> 
> The third part is How to know when to apply the patch?  In pseudo-codish 
> form, here is what I can come up with.
<snip code on applying patch>
> That should go into pkg_postinst, and probably include a sleep delay 
> timer + bell dinging like on portage and baselayout ebuilds to get the 
> attention of users.

I've already got code prepared to take advantage of /usr/src/linux.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[C. Brewer]

[-- Attachment #1: signed data --]
[-- Type: text/plain, Size: 1069 bytes --]

Just a couple of questions, out of curiousity.
1) When is the /usr/src/linux sym going to go away?
2) (snip from Kumba's mail)
                 echo "${BAD}*${NORMAL} benefits.  Please note that this patch 
is only 
for 2.4 kernels.  If you move to a 2.6"
                echo "${BAD}*${NORMAL} kernel, you WILL need to re-merge 
Xfree, 
otherwise OpenGL-based applications"
                echo "${BAD}*${NORMAL} will fail in X."
(snip)
Since this isn't completely accurate..maybe a slight tweaking on the wording.
If you are running a 2.6 kernel with 2.6 headers installed, Open-GL apps will 
fail with X.If you are running a 2.6 kernel with the 2.4 headers installed, 
you won't even notice any change, not to mention avoiding quite a few bs 
problems with packages that wont build against the 2.6 headers anyway.
(I've run xfree 4.3.0-r3 since it hit unstable, and have yet to rebuild it 
through 2.5 to todays 2.6.0test8)

-- 
Chuck Brewer
Registered Linux User #284015
Get my gpg public key at pgp.mit.edu!! Encrypted e-mail preferred.



[-- Attachment #2: signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Luke-Jr]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 19 October 2003 11:48 pm, C. Brewer wrote:
> Since this isn't completely accurate..maybe a slight tweaking on the
> wording. If you are running a 2.6 kernel with 2.6 headers installed,
> Open-GL apps will fail with X.If you are running a 2.6 kernel with the 2.4
> headers installed, you won't even notice any change, not to mention
> avoiding quite a few bs problems with packages that wont build against the
> 2.6 headers anyway. (I've run xfree 4.3.0-r3 since it hit unstable, and
> have yet to rebuild it through 2.5 to todays 2.6.0test8)
Is this an issue at all then? Gentoo doesn't have a 2.6 linux-headers...
- -- 
Luke-Jr
Developer, Gentoo Linux
http://www.gentoo.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQE/k4sAZl/BHdU+lYMRAksgAKCS15nIbis5t8IvPVoa41yh+zWYzwCgiG3Y
48OpgDAZKUUbnTAR5+epjH4=
=+zAL
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[C. Brewer]

[-- Attachment #1: signed data --]
[-- Type: text/plain, Size: 1243 bytes --]

On Monday 20 October 2003 7:13, Luke-Jr wrote:
> On Sunday 19 October 2003 11:48 pm, C. Brewer wrote:
> > Since this isn't completely accurate..maybe a slight tweaking on the
> > wording. If you are running a 2.6 kernel with 2.6 headers installed,
> > Open-GL apps will fail with X.If you are running a 2.6 kernel with the
> > 2.4 headers installed, you won't even notice any change, not to mention
> > avoiding quite a few bs problems with packages that wont build against
> > the 2.6 headers anyway. (I've run xfree 4.3.0-r3 since it hit unstable,
> > and have yet to rebuild it through 2.5 to todays 2.6.0test8)
>
> Is this an issue at all then? Gentoo doesn't have a 2.6 linux-headers...
> --
> Luke-Jr
> Developer, Gentoo Linux
> http://www.gentoo.org/

It does, but it's not like a simple emerge thing, there's provisions in the 
dev-sources ebuilds for the headers, and really, when you're running a 2.6, 
it's usually assumed you can creat the headers yourself...it's not that 
hard:)

P.S Please dont cc me on replies to the list or vice versa, I haven't set up 
to filter the duplicates.
-- 
Chuck Brewer
Registered Linux User #284015
Get my gpg public key at pgp.mit.edu!! Encrypted e-mail preferred.



[-- Attachment #2: signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Luke-Jr]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 20 October 2003 04:59 am, Donnie Berkholz wrote:
> > First, why check /usr/src/linux?  IMHO, that link may or may not always
> > be there or be correct.  I recommend instead using the get_KV and
> > KV_to_int functions defined in /sbin/functions.sh to determine what the
> > machine is currently running.
>
> Because everything in Gentoo that needs a kernel should compile against
> /usr/src/linux. pcmcia-cs does so, xfree-drm does so, nvidia-kernel does
> so, ati-drivers does so ...
IIRC, get_KV etc all check /usr/src/linux. Using functions to get the version 
would just make it easier if the policies on how to choose a kernel 
changed...
- -- 
Luke-Jr
Developer, Gentoo Linux
http://www.gentoo.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQE/k29eZl/BHdU+lYMRAr3PAJ9VrDfY2EoPTI01ZWm+3n0gFekf4ACff4If
oAl+B/nOde0brzuzBzcamVQ=
=qI/M
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 335 bytes --]

On Mon, 2003-10-20 at 01:15, Luke-Jr wrote:
> IIRC, get_KV etc all check /usr/src/linux. Using functions to get the version 
> would just make it easier if the policies on how to choose a kernel 
> changed...

Kumba already pointed to /sbin/functions.sh, which contains this
snippet:

get_KV() {
    local KV="$(uname -r)"


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Patches and the 2.4->2.6 move

[Alexander Gretencord]

On Monday 20 October 2003 04:34, Donnie Berkholz wrote:
> This means that if a user emerges xfree when linked to a 2.4 kernel,
> that user will need to remerge xfree after moving to 2.6 kernels. This
> takes about 40 minutes on a ~2GHz x86.

Well 40 minutes on 2GHz ist a lot more time on a 1GHz system. I recently built 
gentoo on my laptop (1.1GHz) and it took more than 12 hours for (not all of) 
kde. With X and gcc and whatnot it would have been quite more. Of course 
every package only adds an hour or two but it still sums up. I merged gtk+ 
with -O just to avoid upgrading to xfree-4.3.0-r3 and even that took quite 
some time.

> 1) Is this acceptable?
> 2) If not, what is a better solution? Dropping the patch entirely? Note
> that I don't know Mesa/programming well enough to write a patch
> compatible for both 2.4 and 2.6 kernels.

At least don't put the reason for this into a postinst notice. I'd hate to 
merge a new xfree, just to see a notice (after hours of compiling) about 
breakage. If you break something, tell me _before_ I merge the package.


Alex


--
gentoo-dev@gentoo.org mailing list